W3C home > Mailing lists > Public > public-xml-processing-model-wg@w3.org > July 2006

Re: p:step

From: Alex Milowski <alex@milowski.org>
Date: Thu, 20 Jul 2006 07:59:44 -0700
Message-ID: <44BF9A60.9060505@milowski.org>
To: public-xml-processing-model-wg@w3.org

Norman Walsh wrote:
> Looking more closely at your examples, I find your use of name/from
> *very* confusing:
>    <p:step kind="p:xslt">
>       <p:input name="input" from="request"/>
>       <p:input name="stylesheet">...<p:input>
>       <p:output name="url-request" from="output"/>
>    </p:step>
> In the first case, I think you're saying that the xslt component's
> input named "input" comes from the thing labelled "request" and its
> input named "stylesheet" comes from the inline content. Then you say
> that its output named "output" has the label "url-requiest".

I made them consistent in that:

* 'name' is always the thing that is being bound
* 'from' is always the referenced infoset.

I ran this by some people and they didn't get it right away so this
might not be the right set of names.  I also ran name/label by those
same people, but they found the use of 'name' and 'label' together
to be way more confusing.

> To me that's using "from" in two completely different ways that I find
> really odd.

Actually, 'from' is being use in the same way... it is just that the use
of 'name' is different.  On an input, it refers to an intrinsic name
of the component.  In the case of an output, it refers to the name
being provided throughout the rest of the pipeline.

> I'd prefer to say:
>   The attribute "from" points to a label
>   The attribute "XXX" provides a label
>   The attribute "YYY" identifies the name of the component stream

Maybe we step back an look at what each of these are doing:

    An input binding is:
      * a name of an input declared by the component
      * one of:
          - a reference to an infoset named somewhere in the pipeline
          - an instance of an infoset

   An output binding is:
     * a name of an output declared by the component
     * a name that is provided to the rest of the pipeline for
       that resulting infoset from running the step.

If we can agree on these semantics, which we probably have, then
we can suggest naming and see what kind of consensus we have.

--Alex Milowski
Received on Thursday, 20 July 2006 14:59:58 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:32:40 UTC